Table 2.

Features of the best individual model (Model 1) and the best ensembles selected.

Model/ensemble	AUC ROC curve training set	AUC ROC curve test set	Sp training set*	Se training set*	Overall accuracy training set*	Sp test set*	Se test set*	Overall accuracy test set*
Model 1	0.796	0.748	78.8%	68.3%	74%	63.4%	74%	66%
Ensemble 1	0.850	0.785	83.5%	74.7%	79%	70.4%	74%	71.4%
Ensemble 2	0.902	0.804	84.7%	79.7%	82%	76%	70.4%	74.5%

*Considering zero as a cutoff value between substrates and non-substrates.